sum_model_lr2e_4_20epoch

This model is a fine-tuned version of weny22/sum_model_t5_saved on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.9106
• Rouge1: 0.2122
• Rouge2: 0.084
• Rougel: 0.1728
• Rougelsum: 0.173
• Gen Len: 18.966

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 0.0002
• train_batch_size: 64
• eval_batch_size: 64
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
No log	1.0	335	2.2399	0.1924	0.0637	0.1541	0.1542	18.9313
3.4862	2.0	670	2.1627	0.1955	0.0665	0.1567	0.1569	18.9167
2.5411	3.0	1005	2.0873	0.2007	0.0692	0.16	0.16	18.936
2.5411	4.0	1340	2.0591	0.1987	0.0707	0.1594	0.1594	18.9753
2.3898	5.0	1675	2.0138	0.2007	0.0736	0.1622	0.1622	18.9567
2.3008	6.0	2010	2.0109	0.2037	0.0752	0.1642	0.1641	18.9393
2.3008	7.0	2345	1.9990	0.2028	0.0748	0.1645	0.1646	18.9513
2.231	8.0	2680	1.9738	0.2059	0.078	0.1677	0.1678	18.9573
2.1849	9.0	3015	1.9619	0.2067	0.0792	0.1685	0.1687	18.9433
2.1849	10.0	3350	1.9461	0.2111	0.0827	0.1726	0.1727	18.9567
2.137	11.0	3685	1.9393	0.2092	0.0813	0.1704	0.1706	18.962
2.1086	12.0	4020	1.9273	0.2092	0.0822	0.1701	0.1702	18.9553
2.1086	13.0	4355	1.9320	0.2096	0.0824	0.1701	0.1702	18.9667
2.0801	14.0	4690	1.9234	0.2119	0.0833	0.1723	0.1723	18.9647
2.0584	15.0	5025	1.9153	0.2115	0.0838	0.1729	0.1732	18.9653
2.0584	16.0	5360	1.9139	0.2116	0.0834	0.173	0.1732	18.9567
2.0356	17.0	5695	1.9130	0.2108	0.0834	0.1723	0.1723	18.976
2.0283	18.0	6030	1.9122	0.2113	0.084	0.1724	0.1726	18.9607
2.0283	19.0	6365	1.9117	0.2122	0.0845	0.1727	0.1729	18.9673
2.0149	20.0	6700	1.9106	0.2122	0.084	0.1728	0.173	18.966

Framework versions

• Transformers 4.38.2
• Pytorch 2.2.1+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2